Striking the right balance: Navigating antimicrobial stewardship and antibiotic prescribing after CAR‐T‐cell therapy

In this issue Keri et al.1 retrospectively examined infections in 76 patients treated with standard-of-care chimeric antigen receptor T-cell therapy (CAR-T-cell) for B-cell malignancies, highlighting a high incidence of Clostridioides difficile infection (CDI). Notably, CDI frequently occurred in some patients receiving antibiotics for fever without any microbiologically documented infection. Considering these findings, and the recognized importance of host-microbiota interactions for CAR-T-cell therapy outcomes, the authors underscore the need for enhanced antimicrobial stewardship. Clostridioides difficile was the most common bacterial event in the cohort, affecting 8% of all patients. CDI was diagnosed based on positive glutamate dehydrogenase (GDH) and toxin detection by enzyme immunoassay (EIA) in patients with diarrhea, with only one episode requiring PCR testing due to GDH/EIA discrepancy. This is significant given the frequent occurrence of diarrhea from other causes, which can lead to overdiagnosis of CDI in cases of colonization. Table 1 summarizes CAR-T-cell trials and observational studies reporting CDI outcomes, demonstrating comparable incidence.2-5 Notably, CDI was a common cause of microbiologically confirmed early bacterial infections, but was an uncommon cause of late infections,6 even in patients with enduring neutropenia.7 NHL ALL NHL ALL CLL NHL ALL The frequent use of broad-spectrum antibiotics following CAR-T-cell infusion for fever in neutropenic patients contributes to CDI risk. Antibiotics are frequently continued even in the absence of microbiologically documented infection, as observed in nearly 90% of patients with cytokine release syndrome (CRS) in this study. Other cohorts report receipt of broad-spectrum antibiotics in 75%–86% of patients following lymphodepletion.2, 3 In centers utilizing bacterial primary prophylaxis, escalation from oral quinolones to broad-spectrum antibiotics at onset of fever is also reported in up to 94% of febrile patients.8 There is lack of data exploring the temporal relationship between exposure to broad-spectrum antibiotics and subsequent CDI. Keri et al. reported patients with CDI were exposed to 3–10 days of antibiotics prior to CDI. Similarly, Logue et al.10 reported that 9 of 12 patients with CDI (study cohort = 31) were receiving broad-spectrum antibiotics (n = 8) or prophylaxis (n = 1) at the time of CDI diagnosis, while Baird et al. reported at least 7 days of broad-spectrum antibiotic exposure prior to CDI.4 Balancing the need to prevent infection during febrile neutropenia while minimizing undesirable effects of broad-spectrum antibiotic use, including the risk of C. difficile, remains a challenge for clinicians. Keri et al. call for increased focus on antimicrobial stewardship in CAR-T-cell therapy recipients. A key challenge in this area is the multiple, often concurrent, potential causes of fever following infusion. Melica et al. recently detailed the myriad aetiologies of fever in a large cohort of CAR-T-cell therapy recipients (n = 232). Of patients who developed a fever between lymphodepletion and D30 (76% of cohort), 69% had CRS, 22% had an infection event, and 18% of patients with CRS also had a confirmed infection.11 Evidence-based de-escalation strategies, such as the cessation of antibiotics regardless of neutrophil count after a fever-free period of 72 hours,12 may be difficult to practically implement in CAR-T-cell therapy recipients with ongoing or recurrent fevers. Two studies have reported antimicrobial stewardship (AMS) interventions in CAR-T-cell therapy recipients.13, 14 An unblinded, randomized-controlled trial compared existing practice with an AMS strategy that ceased antibiotics after 48–72 hours of empiric treatment, if there was no concern for active infection.13 In the sub-analysis of CAR-T-cell therapy recipients (n = 24 per group), there was a non-significant (p = 0.092) increase in antibiotic-free days (5.5 vs. 1, intention-to-treat analysis) in stewardship group, with a median duration of neutropenia of 7 days.13 Safety outcomes such as break-through fever, or post-cessation ICU admission, were not reported for the cellular therapy patients specifically, but did not differ significantly in the overall cohort when comparing the stewardship intervention to standard of care.13 Lucena et al. described a more conservative de-escalation strategy where antibiotics were ceased after 72 h of empiric therapy, if the patient remained afebrile for 72 h and there was no clinical concern for infection.14 The program successfully (p = 0.034) reduced median antibiotic exposure from 8-days to 6-days in their cohort (n = 131), which included a modest number of cellular therapy patients (n = 12, 9%).14 However, sub-analysis of antibiotic duration amongst cellular therapy patients was not provided.14 Antimicrobial prophylaxis during neutropenia adds complexity to stewardship and CDI management. Approaches to fluoroquinolone prophylaxis vary across institutions, and evidence for its use in CAR-T-cell therapy cohorts remains limited. The CAR-HEMATOTOX score (HT),15 assessing baseline inflammation and haematopoetic reserve to stratify the risk of hematotoxicity, was also shown to predict bacterial infection after CD19- and BCMA-directed CAR-T therapies.2, 16, 17 Retrospective analysis indicates that fluoroquinolone prophylaxis significantly reduced bacterial infections in patients with high HT scores, but not in lymphoma patients with low HT scores.2 Further, the EASIX and modified EASIX scores use biomarkers of endothelial dysfunction to predict immune-adverse events, including CRS.18, 19 These tools may help clinicians decide on the necessity of prophylactic antibiotics and antimicrobial escalation during fever, although prospective validation of risk-based prophylaxis in CAR-T-cell cohorts is required. However, with approximately 90% of patients developing fever during neutropenia, leading to broad-spectrum antimicrobial use irrespective of prophylaxis, a more pressing question may be how to better distinguish infectious and non-infectious fever syndromes. Various risk stratification tools have been assessed in this context. National Early Warning Scores (NEWS) were higher in patients without infection at the onset of first fever after CAR-T-cell therapy, compared to those with infections.8 A small prospective study (n = 16) of CAR-T-cell-treated patients in the intensive care unitsuggested that a procalcitonin level (PCT) < 0.5 µg/L may be useful in excluding infection,20 supported by a larger retrospective (n = 98) study with an association between infection and PCT ≥ 0.4 µg/L within the first 48 hours of fever.21 Rejeski et al. found that discriminative validity improved if higher PCTs (≥1.5 µg/L) were combined with pre-infusion clinical risk scores.22 The role of serum proteomics,22 and metagenomics,23 in distinguishing infectious from non-infectious causes of fever is promising, and is being actively investigated in the CAR-T-cell therapy setting. The extent to which the host gut microbiota may independently modulate the effect of CAR-T-cell therapies is a major research focus and underscores the need for meaningful stewardship opportunities. Antibiotics are well known to profoundly impact the structure and function of the gut microbiota.24, 25 Although exposure to broad-spectrum antibiotics during the weeks prior to CD19 CAR-T-cell infusion is a clear surrogate for worse functional status and increased systemic inflammation, it has also been associated with worse overall survival, increased toxicity, and progressive disease in two multisite retrospective studies.26, 27 Metagenomic shotgun sequencing of pre-infusion stool samples from a subset of patients within these cohorts was correlated clinical outcomes, demonstrating associations between specific bacterial taxa and metabolic pathways and key post CAR-T-cell outcomes that warrant further investigation.26 Less well described is the impact of the antibiotic exposure on the gut microbiome in the presence of multiple confounders including steroids and other immune-modulating therapies following CAR-T-cell therapy. Further pre-clinical and clinical studies are needed to elucidate the underlying mechanisms of the microbiome interaction with the immune system in the context of CAR-T-cell therapy. Only then can the full impact of antibiotics on the microbiota and the role of the microbiota in modulating therapeutic outcomes be fully understood. The recent American Society for Transplantation and Cellular Therapy28 best-practice statement highlights several areas of research needed to improve supportive care of CAR-T-cell therapy recipients, including microbiome endpoints for antimicrobial stewardship interventions, approaches to infection risk stratification, improved understanding of immune reconstitution, and long-term infection risk. Keri et al. provide an example of the clinical data that drives these recommendations. Their study emphasizes the need for individualized antimicrobial strategies, the importance of rapid diagnostics to reduce unnecessary antibiotic exposure, and highlights CDI as an important complication in this context. Future research should aim to integrate microbiome analysis and risk stratification tools into clinical practice to optimize outcomes for CAR-T-cell therapy recipients. Gemma Reynolds: Conceptualization, data curation, writing - original draft, writing - review & editing. Olivia Smibert: Conceptualization, data curation, writing - original draft, writing - review & editing. Erika Kampouri: Conceptualization, data curation, writing - original draft, writing - review & editing. Data sharing is not applicable—no new data are generated.